Exploratory analysis of multi-element geochemical patterns in soil from the Sarno River Basin (Campania region, southern Italy) through compositional data analysis (CODA)

The Sarno River Basin (south-west Italy), nestled between the Somma-Vesuvius volcanic complex and the limestone formations of the Campania-Apennine Chain, is one of the most polluted river basins in Europe due to widespread industrialization and intensive agriculture. Water from the Sarno River, which is heavily contaminated by the discharge of human and industrial wastes, is partially used for irrigation on the agricultural fields surrounding it. We apply compositional data analysis to 319 soil samples, collected during two field campaigns along the river course and throughout the basin, to determine the concentration and possible origin (anthropogenic and/or geogenic) of the elemental anomalies, including potentially toxic elements (PTEs). The concentrations of 53 elements were determined using ICP-MS and, subsequently, log-transformed. Using hierarchical clustering, clr-biplot and a principal factor analysis, the variability and the correlations between a subset of extracted variables (26 elements) were identified. Factor score interpolated maps were then generated using both lognormal data (NDR) and clr-transformed data to better visualize the distribution and potential sources of the patterns in the Sarno Basin. The underlying geology substrata appear to be associated with raised levels of Na, K, P, Rb, Ba, V, Co, B, Zr, and Li, due to the presence of pyroclastic rocks from Mt Somma-Vesuvius. Similarly, elevated Pb, Zn, Cd, and Hg concentrations are most likely related to the geological and anthropogenic sources, the underlying volcanic rocks, and contamination from fossil fuel combustion associated with nearby urban centers. Interpolated factor score maps and the clr-biplot show a clear correlation between Ni and Cr in samples taken along the Sarno River, and Ca and Mg near the Solofra district After considering nearby anthropogenic sources, the Ni and Cr are PTEs most likely originating from the Solofra tannery industry, while Ca and Mg correlate to the underlying limestone-rich soils of the area. This study shows the applicability of the log-ratio transformations to these studies, as they clearly show relationships and dependencies between elements which can be lost when univariate and classical multivariate analyses are employed on raw and lognormal data.